A “natural” hypothesis is one with high probability. A “natural” question is a query regarding the cause(s) of a low-probability observation.
So, in this exchange, byrnema pointed to a particular low-probability observation (the abundance of causal structure in the world around us), and Eliezer responded by noting that the proposed explanation (a complex first cause) has low probability, even conditioning on the observation.
To put it in even simpler terms: Bayes’s theorem says P(H|E) = P(E|H)P(H)/P(E); byrnema said: “P(E) is small!”; and Eliezer said “Oh yeah, well P(H) is tiny!”
My reduction-proposal:
A “natural” hypothesis is one with high probability. A “natural” question is a query regarding the cause(s) of a low-probability observation.
So, in this exchange, byrnema pointed to a particular low-probability observation (the abundance of causal structure in the world around us), and Eliezer responded by noting that the proposed explanation (a complex first cause) has low probability, even conditioning on the observation.
To put it in even simpler terms: Bayes’s theorem says P(H|E) = P(E|H)P(H)/P(E); byrnema said: “P(E) is small!”; and Eliezer said “Oh yeah, well P(H) is tiny!”